Abstract
The identification of strip steel surface defect is an important index to test its quality. With the development of deep learning technology, strip surface detection has developed from traditional machine learning algorithm recognition to deep neural network detection. Existing deep learning strip detection algorithms mainly focus on the tuning and identification of standard defect data sets, while few focus on the preliminary preparation processes such as the collection and pretreatment of the original data of the production line. To address this problem, this paper proposes a series of image data acquisition and pre-processing algorithms for strip surface defect detection technology, such as out-of-area background shielding, noise filtering, suspected defect image acquisition, light equalization and image enhancement processing. Light equalization and image enhancement processing and a series of image data acquisition and pre-processing algorithms to quickly eliminate invalid data from the collected image data and retain valid data. And for the image ROI region data broadening, effectively solve the problem of insufficient defect samples, for the existing deep learning strip inspection algorithm to provide a standard training and testing data set. The image acquisition pre-processing and data augmentation algorithm in this paper is also of practical value when extended to other products in the field of surface inspection.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
References
Srinivasan, K., Dastoor, P.H., Radhakrishnaiah, P., et al.: FDAS: a knowledge-based framework for analysis of defects in woven textile structures. J. Text. Inst. 83(3), 431–448 (1992)
Cui, Y., Jia, M., Lin, T.Y., et al.: Class-balanced loss based on effective number of samples. In: Proceedings of the 2019 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), Long Beach, CA, USA, 15 June–20 June 2019, pp. 9260–9269 (2019)
Land, E.H.: The retinex theory of color vision. Sci. Am. 237(6), 108–128 (1978)
Wei, W., Shan, S., Wen, G., et al.: An improved active shape model for face alignment. In: Proceedings of the 2002 IEEE International Conference on Multimodal Interfaces (ICMI), Pittsburgh, PA, USA, 16 October 2002, pp. 523–528 (2002)
Singh, K.K., Lee, Y.J.: Hide-and-seek: forcing a network to be meticulous for weakly-supervised object and action localization. In: Proceedings of the 16th IEEE International Conference on Computer Vision (ICCV), Venice, Italy, 22 October–29 October 2017, pp. 3544–3553 (2017)
Chawla, N.V., Bowyer, K.W., Hall, L.O., et al.: SMOTE: synthetic minority over-sampling technique. J. Artif. Intell. Res. 16(1), 321–357 (2002)
Acknowledgements
The authors would like to express appreciation to Shanghai Key Laboratory of Intelligent Manufacturing & Robotics and all members of the CIMS laboratory for their support. Thanks for the funding from Shanghai Science and Technology Committee of China (No. 19511105200).
Author information
Authors and Affiliations
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2022 The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd.
About this paper
Cite this paper
Wan, X., Liu, L., Gao, Z., Zhang, X. (2022). Research on Image Acquisition Preprocessing and Data Augmentation Algorithm Based on Strip Surface Defect Detection Technology. In: Wang, Y., Martinsen, K., Yu, T., Wang, K. (eds) Advanced Manufacturing and Automation XI. IWAMA 2021. Lecture Notes in Electrical Engineering, vol 880. Springer, Singapore. https://doi.org/10.1007/978-981-19-0572-8_16
Download citation
DOI: https://doi.org/10.1007/978-981-19-0572-8_16
Published:
Publisher Name: Springer, Singapore
Print ISBN: 978-981-19-0571-1
Online ISBN: 978-981-19-0572-8
eBook Packages: Intelligent Technologies and RoboticsIntelligent Technologies and Robotics (R0)